Zhiyu Wang | 王智语

Research Scientist @ Sapient Intelligence
MPhil Advanced Computer Science @ Cambridge
BSc Computer Science @ UCL

My work spans agentic systems, large language models, and geometric machine learning, with a particular interest in bringing these approaches together for AI-driven biology and health research.

I am currently a Research Scientist at Sapient Intelligence, where I focus on developing Praxist, a general-purpose autonomous AI-scientist framework designed for long-running research tasks across domains. Praxist is open source; if you find it useful, please give it a star on GitHub! I am also leading a joint project with Shanghai Jiao Tong University that applies Praxist to the Virtual Cell Challenge 2026, and collaborating with Prof. Jian Tang at Mila on Bayesian optimization for auto-research systems.

Before joining Sapient, I was a Master’s student at Cambridge under the supervision of Prof. Pietro Liò, where I focused on AI for proteins. I first-authored PLASMA, an ICLR 2026 paper on protein substructure alignment, and received the Highly Commended M.Phil. Project Prize for my Master’s thesis on using hierarchical hypergraphs to add a secondary-structure level to protein representation learning.

News

Aug 2026	Praxist will be presented at AI4 2026 in Las Vegas—come see us at our booth!
Jan 2026	Our PLASMA paper on protein local alignment via optimal transport has been accepted at ICLR 2026!
Jun 2025	I am delighted to announce that I have been awarded a £350 Churchill College Postgraduate Academic Travel Fund (PAT0062) to attend IJCAI25 and present my GraphAU-Pain paper.
Jun 2025	Our GraphAU-Pain paper has been accepted for presentation at the MiGA workshop at IJCAI25. Looking forward to seeing everyone in Guangzhou, China on August 28th!

Selected Publications

arXiv

Praxist: From Experimental Artifacts to Solution Lineages

Jin Li^*, Ahmed Murtadha^*, Zhiyu Wang^*, and 13 more authors

2026

Abs Paper Bib

@article{li2026praxist,
  title = {Praxist: From Experimental Artifacts to Solution Lineages},
  author = {Li, Jin and Murtadha, Ahmed and Wang, Zhiyu and Chen, Qiwen and Chen, William and Wu, Yifei and Wang, Guan and Siy, Andy L. and Yang, Jiayi and Huang, Mengsha and Li, Wenhao and Liu, Yixuan and Pan, Shuailin and Yuan, Mingli and Song, Sen and Sun, Yuhao},
  year = {2026},
  eprint = {2608.25955},
  archiveprefix = {arXiv},
  primaryclass = {cs.MA},
  github = {sapientinc/PRAXIST},
}

ICLR
Fast and Interpretable Protein Substructure Alignment via Optimal Transport

Zhiyu Wang^*, Bingxin Zhou^*, Jing Wang, and 4 more authors

In International Conference on Learning Representations (ICLR), 2026

Poster

Abs Paper Bib

PLASMA is a module based on optimal transport that can be seamlessly added to any protein representation learning model with minimal training or even no training to enable fast and highly accurate identification of locally similar substructures. In detecting shared substructures (motifs, binding sites, and active sites), PLASMA reaches ROC-AUC 0.96-0.99 versus 0.81-0.93 for the widely used global aligner TM-align at about 50x its speed, and outperforms the state-of-the-art local method EBA at about 3x its speed, with much cleaner alignments.
@inproceedings{wang2026plasma, title = {Fast and Interpretable Protein Substructure Alignment via Optimal Transport}, author = {Wang, Zhiyu and Zhou, Bingxin and Wang, Jing and Tan, Yang and Zhao, Weishu and Li\`o, Pietro and Hong, Liang}, year = {2026}, booktitle = {International Conference on Learning Representations (ICLR)}, eprint = {2510.11752}, archiveprefix = {arXiv}, primaryclass = {q-bio.BM}, github = {ZW471/PLASMA-Protein-Local-Alignment}, note = {Poster} }
arXiv
Topotein: Topological Deep Learning for Protein Representation Learning

Zhiyu Wang, Arian R. Jamasb, Mustafa Hajij, and 3 more authors

2025

Abs Paper Bib

Topotein is a complete SE(3)-equivariant framework that adds a secondary-structure level to protein representations via a hierarchical hypergraph, the Protein Combinatorial Complex (PCC), processed by the Topology-Complete Perceptron Network (TCPNet). This coarser-grained view gives a better global understanding of protein structure, achieving the best SCOP fold classification accuracy (43.3%, vs. 38.4% for the same network without the hierarchy).
@article{wang2025tcpnet, title = {Topotein: Topological Deep Learning for Protein Representation Learning}, author = {Wang, Zhiyu and Jamasb, Arian R. and Hajij, Mustafa and Morehead, Alex and Braithwaite, Luke and Li\`o, Pietro}, year = {2025}, eprint = {2509.03885}, archiveprefix = {arXiv}, primaryclass = {cs.LG}, github = {ZW471/TopoteinWorkshop} }